腾讯混元大模型中文能力追平GPT-4，16秒视频生成技术获突破
腾讯集团副总裁蒋杰近日在腾讯云生成式AI产业应用峰会上宣布，经过不断的迭代优化，腾讯混元大模型整体性能已跻身国内第一梯队，部分中文能力甚至追平了领先的AI模型GPT-4。这一成就标志着中国本土AI技术在语言理解和生成能力上取得重要进展。
自OpenAI的Sora模型发布以来，AI生成视频技术备受业界关注。腾讯混元大模型同样展现出强大的视频生成能力，支持多种视频生成方式，包括文生视频、图生视频、图文生视频和视频生视频。据透露，腾讯混元已经实现16秒内生成视频的快速响应，这一速度的提升无疑将大大加快视频内容的创作和分发效率。
此外，在3D模型生成领域，腾讯混元也取得了显著进展。模型能够实现文/图生3D，单图30秒内即可生成3D模型，这一速度的提升为3D内容的创作提供了极大的便利，也将为相关行业带来更多创新应用。
腾讯混元大模型的这些突破，不仅展示了腾讯在AI领域的深厚技术积累，也为国内AI产业发展提供了新的动力。随着技术的不断成熟和应用场景的不断拓展，AI技术有望在未来的内容创作、媒体传播、虚拟现实等多个领域发挥更大的作用。
英语如下：
Title: “Tencent’s Harmony Model Matches GPT-4 in Chinese Capabilities, Achieves Breakthrough in 16-Second Video Generation”
Keywords: Harmony Parallel, Video Generation, 3D Model
Content: Tencent’s Harmony model matches GPT-4 in Chinese capabilities, achieves breakthrough in 16-second video generation technology
At a summit on the industrial application of generative AI in Tencent Cloud, Vice President of Tencent Jiang Jie recently announced that after continuous iteration and optimization, the overall performance of Tencent’s Harmony model has entered the first tier in China, with some Chinese capabilities even matching leading AI models like GPT-4. This achievement signifies an important progress in China’s local AI technology in terms of language understanding and generation capabilities.
Since the release of OpenAI’s Sora model, the technology for AI-generated videos has garnered significant industry attention. Tencent’s Harmony model also demonstrates strong capabilities in video generation, supporting various video generation methods including text-to-video, image-to-video, text-and-image-to-video, and video-to-video. It has been revealed that Harmony can now generate videos in just 16 seconds, an improvement that will undoubtedly speed up the creation and distribution of video content.
In the field of 3D model generation, Tencent’s Harmony has also made significant progress. The model can generate 3D models from text/images, producing a 3D model from a single image in just 30 seconds. This speed boost greatly facilitates the creation of 3D content and will bring more innovative applications to relevant industries.
The breakthroughs of Tencent’s Harmony model not only showcase Tencent’s profound technical accumulation in the AI field but also provide new impetus for the development of AI industries in China. As the technology matures and application scenarios expand, AI technology is expected to play a greater role in future content creation, media dissemination, virtual reality, and other fields.
【来源】https://tech.caijing.com.cn/20240517/5012283.shtml
Views: 2